Driverless Tech Must Ride Out the Storms to Come

Navigation products are being released all the time promising drivers the ability to take into account extreme weather conditions.

Matthew Avery, director of insurance research at Thatcham Research, reminds us that drivers need to be aware of the weather at all times in order to ensure that they drive safely. Connectivity can permit them to know what’s ahead of them, allowing them to, for example, adjust their speed and maintain enough distance between them and other vehicles to account for increased braking distance in rainy, snowy and icy road conditions.

He comments: “Firstly, people need to slow down because of reduced visibility but sensor visibility (both camera and radar) have compromised visibility as well, they suffer just like the driver, therefore they need more space and room to react.

“Advanced driver assistance systems (ADAS) currently use either radars or cameras to monitor the environment, e.g. traffic and pedestrians, and react accordingly. These systems can then automatically brake the vehicle (AEB) to help avoid or mitigate a collision.”

At Level 5 of autonomy, the same principle applies except it’s the vehicle itself that needs to be aware of its surroundings to be able to adapt its driving appropriately to the weather conditions it faces at any given moment. Safety, after all, whether human-driven or autonomous is of paramount concern for the autonomous vehicle and its passengers.

Varying development

Saarinen Jari, CTO of full stack autonomous driving software developer for all weather conditions Sensible 4, reveals that autonomous vehicles are currently at various stages of development around the world. Not all of the developers are aiming for Level 5 autonomy: some are more focused on remotely assisting Level 4 autonomous vehicles on pre-defined routes.

He explains: “This sort of Level 4 approach appears to be one major trend, as more and more technology providers find Level 5 too difficult. We can also see the sensor technology taking major leaps forward. The LiDAR’s are becoming more affordable while they improve their measuring performance. The methods to use existing automotive radars improve. We see new 5G technologies emerging, improving the connectivity of the vehicles and enabling new applications.”

His company’s commercial partner, Christian Bering Pedersen CEO of Holo, adds that there has been an expectation that autonomous vehicle technology would evolve much faster than it has done. In fact, he cites the Gartner Hypecycle for 2021 regarding connected and autonomous vehicles, which shows CAVs “at the bottom of the curve and this is where we are at the moment”.

So, rather than being totally independent of human control, his company remotely assists vehicles, which requires human back-up in autonomous driving mode. The ski area outside of Oslo is being used as a 5G test area as the development of faster connectivity is also a crucial factor in their development and adoption. For now, there is a need to deploy vehicles that can perform within the transport system that’s already available. This includes using them on the open road at higher speeds.

Bering Pederson explains why: “This is partly to get better use of the vehicles but just owing to the safety aspect of not following the existing traffic patterns. The usability relates to getting vehicles that can perform in different scenarios. At the moment we don’t have that many vehicles that are usable in different scenarios.”

AI and machine learning

Neil Cawse, CEO of Geotab adds his perspective saying the ability for the connected and autonomous vehicles to detect poor weather or road conditions requires them to use artificial intelligence (AI) and machine learning by analysing camera, radar, LiDAR and other kinds of sensor data.

“To accurately assist CAVs, algorithms play an integral part in dictating a vehicle’s driving decisions while simultaneously identifying when the CAV is in a good position to make a decision and when it is not. For example, if a vehicle is equipped with a video telematics solution such as a dash camera, the camera can become obstructed by dirt or rain. If the vehicle cannot safely drive owing to the weather obstruction, machine learning would provide real time feedback to the system instructing the vehicle to pull over safely. Furthermore, in order to help predict harsh weather reports and conditions, autonomous vehicles can utilize crowd sourced data to gather hyper local weather conditions from the internet and transmit them to the CAV. This can be used to predict when it is not safe for a CAV to drive.”

Anticipating collisions

He thinks that advanced navigation technologies enable autonomous vehicles to anticipate collisions without having to rely on machine learning alone to identify unsafe conditions. “This additional data stream can help make autonomous driving a reality sooner than previously anticipated,” he adds.

However, in his opinion, a number of steps must be taken to enable customers to feel comfortable with adopting autonomous vehicles in their fleets. This requires the development of smart roads alongside CAVs themselves. These smart highways need to be precisely related to their exact mapping, considering the width of the lanes, the ability to recognise a passing or regulator lane, and so on.

“California is in the process of transitioning their roads to thicker lane lines for better visibility for autonomous vehicles,” he reveals before adding: “Changes regarding appropriate regulations for self-driving cars and infrastructure changes such as precise road mapping, are key for customers to feel confident about autonomous vehicles.”

Overcoming concerns

Despite the fact that the key benefits of connected and autonomous vehicles are considered to be safety and convenience, there is still some concern about the idea of being driven autonomously – of not having full human control of the vehicle. Yet Bering Pederson offers some reassurance: “All the vehicles we use are constantly connected and supervised meaning there is always someone making sure the vehicle is performing correctly. At the same time, these vehicles are programmed to stop for everything and are looking in all directions at the same time, and they can react faster than a human owing to their sensors and programming. The vehicles are convenient to use due to the fact that you don’t have to drive them yourself, it will be much more relaxing to drive and at some point, you can even do other things while you drive to work.”

Jari adds: “The autonomous driving technology can be installed not only on the vehicle but also on the infrastructure, for example to complex intersections. The methods developed for vehicles can be used with fixed infrastructure, to create centralized situational awareness. This status data can be used over fast and reliable connectivity at any vehicle to make the traffic safer.”

Increasing efficiency and reliability

Cawse concludes by predicting that within the next five to 10 years, autonomous vehicle technology will have the capacity to collect more data, enabling the CAVs to become more efficient and reliable. This will enable the automotive industry to attract customers who are interested in taking advantage of this technology. For now, there is a need to continue to advance connected vehicle technologies to help provide a safer, more efficient and autonomous future no matter the weather.